新的有偏扩散模型LMC仿真算法的精度（英文）
发布时间：2018-08-30 20:17
【摘要】：已证实传统定时间步长LMC算法,能准确复现粒子分布的均值和方差。而存在漂移时,总产生非零的偏度。为进一步提高精度,设计了具有5个转移概率的新LMC有偏扩散模型算法,且新算法能准确地复现分布的前三阶矩。一旦给定漂移速度和扩散率,有效网格和时间步长的范围就能够数值计算出来。引入了一种新的度量方法来研究新算法的精度特性。从而发现了一个特殊的网格步长,能让算法仿真的首步误差是最大误差。在该网格步长下,也能发现最优时间步长,使得最大误差达到最小。
[Abstract]:It has been proved that the traditional fixed time step size LMC algorithm can accurately reproduce the mean and variance of particle distribution. And when there is drift, there is always a non-zero bias. In order to further improve the accuracy, a new LMC biased diffusion model algorithm with five transition probabilities is designed, and the first three moments of the distribution can be accurately reproduced by the new algorithm. Once the drift velocity and diffusivity are given, the range of effective mesh and time step can be calculated numerically. A new measurement method is introduced to study the accuracy of the new algorithm. Thus, a special mesh step size is found, and the first step error of the algorithm simulation is the maximum error. Under the grid step, the optimal time step can also be found and the maximum error is minimized.
【作者单位】： 国防科学技术大学信息系统与管理学院;
【基金】：National Natural Science Foundation of China(61673388)
【分类号】：TP391.9
，
本文编号：2214172
[Abstract]:It has been proved that the traditional fixed time step size LMC algorithm can accurately reproduce the mean and variance of particle distribution. And when there is drift, there is always a non-zero bias. In order to further improve the accuracy, a new LMC biased diffusion model algorithm with five transition probabilities is designed, and the first three moments of the distribution can be accurately reproduced by the new algorithm. Once the drift velocity and diffusivity are given, the range of effective mesh and time step can be calculated numerically. A new measurement method is introduced to study the accuracy of the new algorithm. Thus, a special mesh step size is found, and the first step error of the algorithm simulation is the maximum error. Under the grid step, the optimal time step can also be found and the maximum error is minimized.
【作者单位】： 国防科学技术大学信息系统与管理学院;
【基金】：National Natural Science Foundation of China(61673388)
【分类号】：TP391.9
，
本文编号：2214172
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2214172.html
